JP2006180267A - Picture quality correcting circuit - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a picture quality correcting circuit generating a correction curve corresponding to a kind of an image by roughly deciding the image. <P>SOLUTION: The picture quality correcting circuit is equipped with: a pixel sampling processing portion which extracts one pixel for every N (N: a natural number) pixels as to a luminance signal Y; a distribution detection portion which comprises a plurality of counters and counts up according to luminance levels of the pixels extracted by the pixel sampling processing portion to detect a histogram distribution by counting; a correction curve generation section generating a correction curve of the luminance signal Y based on this histogram distribution; and a gradation characteristic correction portion which corrects gradation characteristics by using the generated correction curve. The circuit is also is provided with: a spatial frequency detection portion which detects a spatial frequency; and a counter selection portion which selects counters. The correction curve generation portion is equipped with a limitation processing portion which adjusts a counted value from a counter selected by the counter selection portion and where pixels having low spatial frequency values are concentrated. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image quality correction circuit that has been used when displaying images on a display device such as a plasma display panel (PDP), and generates and outputs an optimal correction curve based on a histogram distribution. It relates to the circuit.

Conventionally, when displaying an image on a display device such as a plasma display panel (PDP), image quality correction processing for correcting input / output characteristics based on an input signal has been performed. A correction circuit that detects a distribution of input levels of a video signal, a so-called histogram, selects a correction curve that emphasizes contrast correction in a luminance level portion where the histogram is concentrated, and has already been proposed as Patent Document 1, for example. Has been. Further, since this Patent Document 1 cannot cope with a case where there are two histogram distribution peaks, even in such a case, a correction circuit capable of generating a correction curve for improving the contrast of each peak portion together. This applicant has proposed (Japanese Patent Application No. 2003-310596).
JP 2000-287104 A

  However, even if a correction curve is generated in accordance with the histogram distribution, an optimal image is not always obtained. For example, if a bright background such as a blue sky occupies a large area on the screen and a histogram distribution (when the luminance level is 256 gradations) is taken for an image on which a subject such as a person is displayed, As shown in FIG. 6, since the blue sky in the background is bright, the largest peak appears in the high luminance portion, and the human skin color of the person appears as a smaller peak in the middle luminance portion. When a correction curve is generated based on the histogram distribution shown in FIG. 6, a correction curve as shown in FIG. 7 is obtained. As can be seen from FIG. 7, the high-intensity part of the background with the most concentrated luminance distribution and the middle-intensity part of the subject with the next concentrated luminance distribution are sequentially curved curves, each before correction. Contrast is improved compared to, but as a result of assigning a large number of gradations to the high luminance part of the background, other image parts including the subject are darker as a whole compared to the through characteristics before correction. There was a problem.

  In addition, it is the subject that most wants to improve the contrast, and especially for changes in the skin color of the person, the user will be sensitive to even slight differences, so it will not be dark compared to the through characteristics. In addition, it is necessary to improve the contrast. On the other hand, the effect of improving the contrast is less for backgrounds such as the blue sky, but rather, the demerit that assigning many gradations to the blue sky portion darkens other image portions. Therefore, it is ideal to determine what the subject or background is and then determine whether or not to apply the correction. However, the image cannot be determined only by the histogram distribution. On the other hand, there is a technology that divides an image into regions, specifies the type of image in the region from contour information, color information, etc., and performs image quality processing suitable for the image, but real-time characteristics suitable for moving images And it did not meet the needs for various video inputs.

  The present invention has been made in view of the above-described problems, and performs rough image discrimination while satisfying conditions such as real-time characteristics suitable for moving images and support for various video inputs, and supports the types of images. An object of the present invention is to provide an image quality correction circuit capable of generating a corrected curve.

  Claim 1 of the present invention comprises a pixel sampling processing unit that extracts one pixel from N (N is a natural number) pixels of the input luminance signal Y, and a plurality of counters. Using a distribution detection unit that counts according to the luminance level of the extracted pixel and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and the generated correction curve In the image quality correction circuit comprising a gradation characteristic correction unit for correcting gradation characteristics, a spatial frequency detection unit for detecting a spatial frequency and a counter selection unit for selecting the counter are provided, and the correction curve generation unit Comprises a limit processing unit for adjusting a count value from a counter in which the spatial frequency selected by the counter selection unit is concentrated on pixels of a specific value. It is an image quality correction circuit according to claim Rukoto.

  According to a second aspect of the present invention, in addition to the first aspect, the restriction processing unit adjusts the count value from a counter in which pixels having low spatial frequency values selected by the counter selection unit are concentrated. And an image quality correction circuit.

  According to a third aspect of the present invention, a pixel sampling processing unit that extracts one pixel from N (N is a natural number) pixels of the input luminance signal Y, and a plurality of counters, the pixel sampling processing unit includes: Using a distribution detection unit that counts according to the luminance level of the extracted pixel and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and the generated correction curve In an image quality correction circuit comprising a gradation characteristic correction unit for correcting gradation characteristics, a spatial frequency detection unit for detecting a spatial frequency, the presence of a specific color from the color difference signals Cr and Cb and the value of the spatial frequency. A specific color detection unit that detects the counter and a counter selection unit that selects the counter; and the correction curve generation unit includes the specific color selected by the counter selection unit. It comprises a restriction processing unit that adjusts the count value from the medium to counter a quality correction circuit characterized by comprising.

  According to a fourth aspect of the present invention, in addition to the third aspect, the restriction processing unit performs an adjustment so as to lower a count value from a counter in which skin color as a specific color selected by the counter selection unit is concentrated. The image quality correction circuit characterized by the above.

  Claim 5 of the present invention comprises a pixel sampling processing unit that extracts one pixel for N (N is a natural number) pixels of the input luminance signal Y, and a plurality of counters. Using a distribution detection unit that counts according to the luminance level of the extracted pixel and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and the generated correction curve In an image quality correction circuit comprising a gradation characteristic correction unit for correcting gradation characteristics, a spatial frequency detection unit for detecting a spatial frequency, and a specific color for detecting the presence of a specific color from the values of the color difference signals Cr and Cb A correction unit that includes a detection unit, and a counter selection unit that selects a counter in which the spatial frequency is concentrated on pixels having a specific value and a counter in which the specific color is concentrated Comprises a limit processing unit that is selected as a counter in which the spatial frequency is concentrated on pixels of a specific value by the counter selection unit and adjusts the count value from the counter selected as a counter in which a specific color is concentrated. The image quality correction circuit characterized by the above.

  According to a sixth aspect of the present invention, in addition to the fifth aspect, in the restriction processing unit, the counter selection unit selects the counter having a low spatial frequency value as a concentrated counter and concentrates the skin color as a specific color. The image quality correction circuit is characterized in that adjustment is performed so as to decrease the count value from the counter selected as.

  According to a seventh aspect of the present invention, in addition to the first, third, or fifth aspect, a matrix conversion processing unit is provided after the gradation characteristic correction unit, and in the matrix conversion processing unit, the input luminance signal Y and the gradation are provided. The correction ratio is detected from the corrected luminance signal Y ′ from the characteristic correction unit, and the color difference signals Cr and Cb are corrected at the same ratio as the correction ratio, respectively, and then the corrected luminance signal Y ′ and the corrected luminance signal Y ′ are corrected. An image quality correction circuit characterized in that matrix conversion processing is performed using color difference signals Cr and Cb.

  According to an eighth aspect of the present invention, in addition to the first, third, or fifth aspect, a matrix conversion processing unit and a γ correction processing unit are sequentially provided after the gradation characteristic correction unit, and among these, the γ correction processing unit includes the space. When detecting an image with a high spatial frequency and high gradation detected by the frequency detection unit, a γ correction curve is selected so that the high gradation part is not crushed, and γ correction processing is performed. This is an image quality correction circuit.

  According to the first aspect of the present invention, the limit processing unit generates the correction curve after adjusting the count value from the counter in which the spatial frequency selected by the counter selection unit is concentrated on the pixels having a specific value. When the adjustment is made to decrease the count value, the image portion with a low spatial frequency becomes a gentle correction curve. Conversely, when the adjustment is made to increase the count value, the image portion with a high spatial frequency has an inclination. It becomes a correction curve.

  According to the second aspect of the invention, the limit processing unit generates the correction curve after performing the adjustment to reduce the count value from the counter in which the pixels having the low spatial frequency values selected by the counter selection unit are concentrated. Therefore, the flat image portion becomes a gentle correction curve, and as a result, it is possible to generate a correction curve having an inclination in the image portion where the contrast is desired to be improved.

  According to the third aspect of the present invention, the correction curve is generated after adjusting the count value from the counter where the specific color is concentrated. Therefore, when the adjustment is performed to decrease the count value, the image where the specific color is concentrated. On the other hand, when the adjustment is performed to increase the count value, the contrast of the image portion where the specific color is concentrated is improved.

  According to the invention described in claim 4, since the correction curve is generated after the adjustment of increasing the count value from the counter in which the skin color as the specific color is concentrated in the restriction processing unit, the skin color is detected. Since the count value of the portion increases, a correction curve that does not darken the skin color can be generated.

  According to a fifth aspect of the present invention, in the restriction processing unit, the counter selection unit selects a counter from which the spatial frequency is selected as a counter concentrated on a pixel having a specific value and a counter selected as a counter on which a specific color is concentrated. Since the correction curve is generated after adjusting the value, it is possible to detect the specific color more accurately by adding the spatial frequency condition, and when the adjustment is performed to lower the count value of the part corresponding to the detection Is a correction curve in which the contrast of the image portion where the specific color is concentrated is lowered, and conversely, when the adjustment is made to increase the count value, the correction curve is improved in the contrast of the image portion where the specific color is concentrated.

  According to the sixth aspect of the present invention, the count value from the counter selected by the counter selection unit as a counter in which pixels having low spatial frequency values are concentrated and a counter in which specific colors are concentrated is selected by the counter selection unit. Since the correction curve is generated after performing adjustment to increase the skin frequency, it is possible to detect the skin color as a specific color more accurately by adding the spatial frequency condition, and thereby the count value of the part corresponding to the detection Therefore, a correction curve that does not darken the skin color can be generated.

  According to the seventh aspect of the present invention, the correction ratio is detected from the input luminance signal Y and the corrected luminance signal Y ′ from the gradation characteristic correction section, and the color difference signal Cr is detected at the same ratio as this correction ratio. , Cb is corrected, and then the matrix conversion process is performed using the corrected luminance signal Y ′ and the corrected color difference signals Cr and Cb, so that a stable image can be obtained without changing color shades. Can be obtained.

  According to the eighth aspect of the present invention, when an image having a high spatial frequency and high gradation detected by the spatial frequency detection unit is detected, a γ correction curve is selected so that the high gradation portion is not crushed and γ is selected. Since the correction process is performed, the high gradation portion is not crushed even if the γ correction process is performed.

The image quality correction circuit according to the present invention includes a pixel sampling processing unit that extracts one pixel from N (N is a natural number) pixels of the input luminance signal Y, and a plurality of counters. Using a distribution detection unit that counts according to the luminance level of the extracted pixel to detect a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and the generated correction curve In a picture quality correction circuit comprising a tone characteristic correction unit for correcting tone characteristics, a spatial frequency detection unit for detecting a spatial frequency, and a specific color for detecting the presence of a specific color from the values of the color difference signals Cr and Cb A correction unit configured to select a detection unit, a counter in which pixels having a low spatial frequency value are concentrated, and a counter in which the specific color is concentrated; The generation unit includes a restriction processing unit that is selected as a counter in which pixels having a low spatial frequency value are concentrated by the counter selection unit and that adjusts a count value from the counter selected as a counter in which a specific color is concentrated. It is characterized by this.
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  The configuration of the image quality correction circuit according to the present invention will be described with reference to the block diagram of FIG. In FIG. 1, the luminance signal Y from the luminance signal input terminal 11 is input to the pixel sampling processing unit 12 and to the gradation characteristic correction unit 16. The output from the pixel sampling processing unit 12 sequentially passes through the distribution detection unit 13, the correction curve generation unit 14, and the time axis filter unit 15, and then is output as a correction curve to the gradation characteristic correction unit 16, thereby correcting the gradation characteristic. The unit 16 corrects the gradation characteristic of the luminance signal Y based on the correction curve, and outputs it to the matrix conversion processing unit 17. The matrix conversion processing unit 17 performs matrix conversion processing using the corrected luminance signal Y ′ and the color difference signals Cr and Cb from the color difference signal input terminal 29 to obtain R, G, and B signals. The γ correction processing unit 18 performs γ correction and outputs R ′, G ′, and B ′ signals from the output terminal 19. Further, when performing the above correction, a specific image is detected using the average luminance information obtained by the average luminance detection unit 26 and the spatial frequency information obtained by the spatial frequency detection unit 27, and is adjusted to the image. A correction curve is generated.

  The pixel sampling processing unit 12 performs sampling processing for extracting one pixel per N (N is a natural number) for the luminance signal Y from the luminance signal input terminal 11. For example, sampling is performed at a rate of 1 for every 32 pixels, and the result is output to the distribution detector 13 at the subsequent stage. As a result, the scale of the counter can be reduced. Here, the condition of N when extracting one pixel per N may be any. For example, if N = 2, the accuracy of the histogram distribution can be kept high, but the scale of the counter can be increased. If N = 128, the counter scale can be made very small, but the accuracy of the histogram distribution is lowered. Therefore, it is desirable to set an optimal value appropriately.

  In the distribution detector 13, 20 a to 20 p are 16 counters, that is, counter 0 to counter 15, and these 16 counters divide 256 gradations of the luminance signal Y into 16 and each counter has 16 counters. The brightness level for each gradation is counted, and the brightness level of each pixel sampled by the pixel sampling processing unit 12 is counted by a corresponding counter, and the count value of each counter is used as a correction curve generation unit at the subsequent stage. 14 is output to the control processing unit 20. For example, a pixel at the luminance level 18 is counted by the counter 1 and counted by the counter 14 at the luminance level 230.

  The correction curve generation unit 14 includes a limitation processing unit 21 for limiting the output from each counter of the distribution detection unit 13, an integration processing unit 22 for generating a correction curve from the output of the limitation processing unit 21, and a pixel. It comprises a normalization processing unit 23 for normalizing the correction curve obtained from the numbers to units of luminance levels from 0 to 255.

The limit processing unit 21 sets an upper limit and a lower limit for each output of the counter 0 to the counter 15, outputs an upper limit value when the upper limit is exceeded, and outputs a lower limit value when the lower limit is exceeded. Thus, the correction curve is prevented from changing with an extremely steep slope. Further, the limit processing unit 21 performs a process of limiting the upper limit value of the counter based on a selection signal and a flatness rate from a counter selection unit 28 described later. Here, the flatness ratio is a ratio of pixels having a low spatial frequency of the image expressed by a value between 0 and 1. Since the effect of improving the contrast is low for a flat image portion, the flatness ratio is reduced. Accordingly, by limiting the upper limit value of the counter, in the case of a flat image, the correction curve is limited so that the inclination of the correction curve becomes gentle. Specifically, the counter selected by the selection signal from the counter selection unit 28 limits the direction in which the upper limit value decreases by performing the following calculation.
(Upper limit value of arbitrary counter X) = (Count value of counter X) × (1−Flat rate)
According to this calculation, it can be seen that the upper limit is strongly limited in proportion to the value of the flatness ratio.

  The integration processing unit 22 is for generating a correction curve by sequentially integrating the count values from the counter 0 to the counter 15 and the outputs limited by the limitation processing unit 21 respectively. Specifically, starting from 0, first, the count value of counter 0 becomes the first output, and then the count value of counter 0 added to the count value of counter 1 becomes the second output. The count value of the counter 1 is added to the count value of the counter 1 and the count value of the counter 2 is added to the count value of the counter 2 to obtain the third output value. A correction curve is formed by connecting the envelopes of the 16 output values accumulated up to.

  The normalization processing unit 23 is for normalizing the correction curve obtained in units of the number of pixels into units of luminance levels from 0 to 255. In the integration processing unit 22, a correction curve is obtained by performing integration processing in units of the number of pixels counted by the counter 0 to the counter 15, but it is necessary to normalize this to a correction curve in luminance level units. The correction curve generated by the integration processing unit 22 is normalized so as to be expressed by the output luminance level, and is output to the dynamic filter 24 of the time axis filter unit 15 at the subsequent stage.

  The time axis filter unit 15 employs a dynamic filter 24 for suppressing a sudden change in luminance level between frames, and a luminance level before the change when the change in luminance level between frames is a certain value or less. And hysteresis 25. The time base filter unit 15 receives the average luminance level from the average luminance detection unit 26.

  The dynamic filter 24 has a luminance level change of a certain value before and after the switching of the frame in order to prevent a phenomenon in which visibility is lowered due to a sudden change in the luminance level when the frame is switched. In this case, the luminance value is adjusted so as to change gently without being directly changed. Specifically, when the difference between the luminance level a one frame before and the luminance level b of the current frame is not less than the set value m or not more than the set value −m, the brightness level b is not output directly. , (B−a) and the coefficient k (0 <k <1) are added to a and output. That is, when | (b−a) | ≧ m, a + (b−a) × k is output as the luminance level.

  The hysteresis 25 adopts the luminance level before the change without outputting the luminance level of the current frame when the difference between the luminance level of the previous frame and the luminance level of the current frame is a certain value or less. This is, for example, a phenomenon in which an image flickers when a slight change such as a luminance level of 30, 33, 30, 33,... It is intended to prevent this. The constant value can be arbitrarily set. For example, when the luminance level change is 3 or less, if the luminance level before the change is adopted, the luminance levels “30” and “33” as described above. "Is repeated, the luminance level is fixed to" 30 ", and the switching is performed only when the luminance level changes by 4 or more. The output of the hysteresis 25 is input to the subsequent gradation characteristic correction unit 16.

  When processing is performed in the time axis filter unit 15, the response speed of correction becomes slow, and correction not corresponding to the current frame is performed. Therefore, using the average luminance level from the average luminance detection unit 26, the average luminance level of the next frame is compared with the average luminance level of an arbitrary frame (the frame from which the correction curve to be output is generated). If the average luminance level is significantly different, the correction curve from the correction curve generation unit 14 is directly output to the subsequent gradation characteristic correction unit 16 instead of the correction curve processed by the time axis filter unit 15. Thus, it is possible to avoid adverse effects caused by the slow response speed.

  The gradation characteristic correction unit 16 corrects and outputs the gradation characteristic based on the correction curve generated through the distribution detection unit 13, the correction curve generation unit 14, and the time axis filter unit 15 in sequence. Since the correction curve represents the relationship of the output luminance level with respect to the input luminance level, matrix conversion is performed by selecting the output luminance signal Y ′ for the luminance signal Y from the luminance signal input terminal 11 based on the correction curve. Output to the processing unit 17.

  The matrix conversion processing unit 17 receives the corrected luminance signal Y ′ and the color difference signals Cr and Cb from the color difference signal input terminal 29, and uses these signals Y ′, Cr and Cb to generate R, G, A B signal is generated and output to the γ correction processing unit 18. The R, G, and B signals reflect the information of the luminance signal Y ′ whose tone characteristics have been corrected based on the correction curve.

  The γ correction processing unit 18 performs γ correction processing based on the R, G, and B signals, and then outputs the R ′, G ′, and B ′ signals from the output terminal 19. With such a configuration, it is possible to generate R ′, G ′, and B ′ signals that reflect the information of the luminance signal Y ′ whose tone characteristics are corrected and that have undergone γ correction processing. The R ′, G ′, and B ′ signals generated in this way can solve the problem of image quality deterioration in the low gradation portion and the high gradation portion that occurs when only γ correction processing is performed, and obtain a smooth image quality. Can do.

  The average luminance detecting unit 26 calculates an average luminance level for one frame for the input luminance signal Y and outputs the average luminance level to the dynamic filter 15.

The spatial frequency detection unit 27 obtains the spatial frequency for each pixel using the input luminance signal Y. As a specific method, the spatial frequency detection unit 27 obtains the spatial frequency from the luminance difference between the pixel of interest and the adjacent pixel, or Fourier transform. It is calculated by processing. Next, the spatial frequency thus obtained is compared with a threshold value for each pixel, and a pixel having a spatial frequency smaller than the threshold value is determined to be a flat pixel. Then, the flat rate is obtained by performing the following calculation from the ratio between the number of pixels determined to be a flat pixel and the total number of pixels, and is output to the counter selection unit 28 at the subsequent stage.
(Flat rate) = (coefficient) × (number of flat pixels / total number of pixels)
The coefficient takes a value between 0 and 1, and the closer to 1, the stronger the limit to the upper limit value of the counter.

  The counter selection unit 28 calculates the average of the luminance values of the pixels determined to be flat pixels by the spatial frequency detection unit 27, and counts the average luminance value of the flat pixels. A selection signal is output to the restriction processing unit 21 so as to select as a counter for limiting the upper limit value. In addition, the flatness value obtained by the spatial frequency detection unit 27 is also output to the restriction processing unit 21. Note that only one counter may be selected as the counter selected by the counter selection unit 28, or a plurality of counters near the average luminance value may be selected.

Next, the operation of the present invention in the above configuration will be described.
The luminance signal Y input from the luminance signal input terminal 11 is input to the pixel sampling processing unit 12, where sampling processing for extracting one pixel per 32 pixels is performed, and the luminance signal Y of the sampled pixel is obtained. The information is output to the downstream distribution detection unit 13. The luminance signal Y input to the distribution detection unit 13 is counted by any one of the counters 0 to 15 according to the luminance level, counted for one frame, and then output to the restriction processing unit 21 at the subsequent stage. For example, if an image in which a large portion of the image occupies a blue sky as a background and a person appears as an object in the background is input, the blue sky is very high in brightness, so Since the human skin of a person has medium luminance, the counter is counted by the counters 7 and 8. Hereinafter, a case where the entire histogram distribution detected by the distribution detection unit 13 at this time is as shown in FIG. 6 will be described assuming that such an image has been input.

  At the same time, the spatial frequency detection unit 27 detects the spatial frequency of each pixel and obtains the flat rate of the current frame, and the counter selection unit 28 in the subsequent stage obtains the average of the luminance values of the pixels determined as the flat pixels. For example, since the blue sky in the background is an image with a low spatial frequency, the flatness ratio can be said to be a very large value in an image in which the blue sky occupies most. Further, since the blue sky occupies most of the flat pixels, the average luminance value of the flat pixels obtained by the counter selection unit 28 is also a high value, so that the counters 14 and 15 are selected.

The limit processing unit 21 performs an operation for limiting the count value of the counter indicated by the selection signal from the counter selection unit 28 among the count values of each counter input from the distribution detection unit 13. For example, when the counters 14 and 15 are instructed by a selection signal,
(Upper limit value of counter 14) = (Count value of counter 14) × (1−Flat rate)
(Upper limit value of counter 15) = (Count value of counter 15) × (1−Flat rate)
Is performed to limit the count value of each counter. When the flatness is a very large value, the count value of each counter is greatly limited as shown in FIG. The count value after performing the limiting process in this way is output to the subsequent integration processing unit 22.

  Thereafter, a correction curve is generated by sequentially integrating the count value after the limit processing in the integration processing unit 22, and the correction curve obtained in units of the number of pixels in the normalization processing unit 23 is a luminance of 0 to 255. Normalize to level units. The normalized correction curve is processed by suppressing a sudden change in luminance level between frames by the dynamic filter 24 of the time axis filter unit 15 and suppressing a slight change in luminance level between frames by the hysteresis 25. Are output to the gradation characteristic correction unit 16. The gradation characteristic correction unit 16 corrects and outputs the gradation characteristic based on the generated correction curve, and sequentially outputs the output from the output terminal 19 after performing matrix conversion processing and γ correction processing.

  The correction curve of FIG. 4 according to the present invention and the correction curve of FIG. 7 according to the prior art are generated from the same image. As can be seen from FIG. 4, in the present invention, the background (blue sky) occupying most of the screen. As a result of this, the problem that the entire image becomes dark as a result of being restricted and a gentle curve is solved, and it is not necessary to assign many gradations to the background part, so that the correction curve of the subject part is inclined. It is stronger than 7.

  As described above, the image quality correction circuit of the present invention does not discriminate the type of image in detail, but indirectly determines a flat pixel portion having a thin correction effect such as a background from the average luminance value of the flat pixel, and Since the count value of the flat pixel portion is limited according to the flatness ratio, the flat image portion becomes a gentle correction curve, and as a result, the correction curve having an inclination in the image portion where the contrast is to be improved such as the subject. Can be generated.

  In the first embodiment, since the spatial frequency is used as the image determination material in addition to the histogram distribution, it is possible to limit the correction curve so that a correction curve having an inclination is not generated in a flat pixel portion such as a background. However, the present invention is not limited to this. Furthermore, a specific color may be detected using a color difference signal, and the lower limit value may be limited so that the specific color portion has a contrast. Hereinafter, the configuration of the second embodiment will be described with reference to FIG. 2, but the description of the same reference numerals as those in FIG.

  In FIG. 2, reference numeral 30 denotes a specific color detection unit. The specific color detection unit 30 detects a specific color using the color difference signals Cr and Cb from the color difference signal input terminal 29. Here, the specific color is, for example, the skin color of a person, and the skin color is a color that the user sensitively identifies even a slight difference, so the skin color is discriminated and the lower limit value of the corresponding counter is limited. To prevent the skin color from becoming dark.

  As shown in FIG. 5, each color has its own characteristic depending on the type of image. For example, in the human skin region, the Cb component has a small value and a narrow width distribution, but the Cr component is widely distributed from a small value to an intermediate region. In addition, the spatial frequency F is small. In contrast, in the forest area, both Cr and Cb components are distributed in a relatively narrow width, but the spatial frequency F distribution is widely distributed from a small value to an intermediate area. Yes.

Therefore, using this feature, the specific color detection unit 30 detects human skin as an example of the specific color. The detection conditions are as follows.
(1) Cr _min <Cr <Cr _max and Cb _min <Cb <Cb _max
(2) F <F _max
(3) Input luminance level range 64 to 160
_{However, Cr min, Cr max, Cb} min, Cb max, F max denotes the maximum value or the minimum value of the specified region. Among these conditions, (1) is an essential condition, and detection accuracy is improved by sequentially adding the conditions (2) and (3) below.

When the human skin pixels are determined under the above conditions, the specific color detection unit 30 further calculates the human skin rate from the ratio of the human skin pixels to all the pixels by the following calculation.
(Human skin rate) = (coefficient) × (number of pixels of human skin / total number of pixels)
The coefficient takes a value between 0 and 1, and the closer to 1, the stronger the limit to the lower limit value of the counter. The human skin rate obtained in this way is output to the counter selection unit 28.

  In addition to the selection of the counter for limiting the upper limit value performed in the first embodiment, the counter selection unit 28 calculates the average of the luminance values of the pixels determined as human skin pixels in the specific color detection unit 30. Thus, the selection signal is output to the limit processing unit 21 so that the counter of the distribution detection unit 13 counting the average luminance value of the human skin pixel is selected as a counter for limiting the lower limit value. Further, the human skin rate value obtained by the specific color detection unit 30 is also output to the restriction processing unit 21. Note that only one counter may be selected as the counter selected by the counter selection unit 28, or a plurality of counters near the average luminance value may be selected.

In response to this, the limit processing unit 21 limits the lower limit value of the counter value indicated by the selection signal from the counter selection unit 28 among the count values of each counter input from the distribution detection unit 13. The following calculation is performed.
(Lower limit value of arbitrary counter X) = (Count value of counter X) × (1 + human skin rate)
By limiting the lower limit value in this way, if the count value is small despite the presence of human skin pixels, conventionally, if the correction curve is generated as it is, the skin color becomes darker. However, the correction curve that improves the contrast can be generated, and the problem that the human skin becomes dark can be solved.

  In the first and second embodiments, the count value of the counter is limited based on the detection result of the flat pixel detection or the human skin pixel detection, and this is reflected in the correction curve. This is for generating a correction curve for correcting the signal Y, and the color difference signals Cr and Cb are not corrected. However, when only the luminance signal Y is corrected, there is a risk that the shade of the color changes, resulting in a light image or a dark image.

Therefore, the matrix conversion processing unit 17 performs the following calculation on each of the color difference signals Cr and Cb so that the color difference signal is also corrected at the same rate as the luminance signal correction rate.
(Output color difference signal) = (Output luminance signal) / (Input luminance signal) × (Input color difference signal)
By this calculation, correction is applied to each of the color difference signals Cr and Cb at the same rate as the change of the luminance signal, so that a stable image can be obtained without changing the color shade.

Further, not only the correction at the same rate as the change of the luminance signal but also the color difference signal may be corrected positively. In the specific color detection unit 30, by combining conditions such as a specific flat rate, a specific luminance, and a specific color region, it is possible to detect to some extent that the image is a specific image. When a specific image is detected, the detection result is output from the specific color detection unit 30 to the matrix conversion processing unit 17 as shown in FIG. The color difference signal is adjusted by the following calculation according to the ratio of the pixels that have been performed (specific image detection rate).
(Output color difference signal) = (Input color difference signal) × (1 + specific image detection rate)
(Specific image detection rate) = (coefficient) × (number of pixels of specific image / total number of pixels)
By this calculation, for example, it is possible to correct the human skin of the person to be red only in the video in which the person appears without changing the hue in the video without the person.

  In the above embodiment, when the γ correction is performed by the γ correction processing unit 18, an optimal γ correction curve is selected from the average input level. However, in a video with a high spatial frequency and high gradation, a high gradation portion is present. There is a risk that the γ correction curve to be crushed will be selected. Therefore, the spatial frequency detection unit 27 counts the number of pixels having a high spatial frequency and a high gradation, and outputs this to the γ correction processing unit 18 as shown in FIG. The γ correction processing unit 18 may select a γ correction curve that does not collapse the high gradation portion when the image has a high spatial frequency and high gradation.

1 is a block diagram illustrating a configuration of an image quality correction circuit according to the present invention. It is the block diagram which showed the structure of the image quality correction circuit of the other Example by this invention. It is the schematic diagram showing the histogram distribution at the time of restrict | limiting with the restriction | limiting process part 21 of this invention with respect to the histogram distribution of FIG. FIG. 4 is a schematic diagram illustrating a correction curve generated based on the histogram distribution of FIG. 3. It is the schematic diagram showing each characteristic of the human skin area | region and the forest area | region as an example of a specific color. It is the schematic diagram showing the histogram distribution calculated | required about the image which a to-be-photographed object (a person etc.) is reflected on the flat and high-intensity background (occupying most images). FIG. 7 is a schematic diagram illustrating a correction curve generated based on the histogram distribution of FIG. 6.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 ... Luminance signal input terminal, 12 ... Pixel sampling process part, 13 ... Distribution detection part, 14 ... Correction curve generation part, 15 ... Time axis filter part, 16 ... Tone characteristic correction part, 17 ... Matrix conversion process part, 18 Γ correction processing unit, 19 output terminal, 20a to 20p, counter 0 to counter 15, 21 limit processing unit, 22 integration processing unit, 23 normalization processing unit, 24 dynamic filter, 25 hysteresis, 26: average luminance detector, 27 ... spatial frequency detector, 28 ... counter selector, 29 ... color difference signal input terminal, 30 ... specific color detector.

Claims (8)

  The input luminance signal Y includes a pixel sampling processing unit that extracts one pixel for every N pixels (N is a natural number), and a plurality of counters according to the luminance level of the pixel extracted by the pixel sampling processing unit. A distribution detection unit that counts and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and a gradation that corrects gradation characteristics using the generated correction curve An image quality correction circuit comprising a characteristic correction unit includes a spatial frequency detection unit that detects a spatial frequency and a counter selection unit that selects the counter, and the correction curve generation unit is selected by the counter selection unit And a limiting processing unit for adjusting a count value from a counter in which the spatial frequency is concentrated on pixels having a specific value. Quality correction circuit.   2. The image quality correction circuit according to claim 1, wherein the restriction processing unit performs adjustment so as to decrease a count value from a counter in which pixels having low spatial frequency values selected by the counter selection unit are concentrated.   The input luminance signal Y includes a pixel sampling processing unit that extracts one pixel for every N pixels (N is a natural number), and a plurality of counters according to the luminance level of the pixel extracted by the pixel sampling processing unit. A distribution detection unit that counts and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and a gradation that corrects gradation characteristics using the generated correction curve In an image quality correction circuit comprising a characteristic correction unit, a spatial frequency detection unit that detects a spatial frequency, a specific color detection unit that detects the presence of a specific color from the color difference signals Cr and Cb and the value of the spatial frequency, A counter selection unit that selects the counter, and the correction curve generation unit is configured from a counter in which the specific colors selected by the counter selection unit are concentrated. Image quality correction circuit characterized by comprising comprises a limiting unit that adjusts the count value.   4. The image quality correction circuit according to claim 3, wherein the restriction processing unit performs adjustment so as to reduce a count value from a counter in which skin color as a specific color selected by the counter selection unit is concentrated.   The input luminance signal Y includes a pixel sampling processing unit that extracts one pixel for every N pixels (N is a natural number), and a plurality of counters according to the luminance level of the pixel extracted by the pixel sampling processing unit. A distribution detection unit that counts and detects a histogram distribution, a correction curve generation unit that generates a correction curve of the luminance signal Y based on the histogram distribution, and a gradation that corrects gradation characteristics using the generated correction curve In an image quality correction circuit comprising a characteristic correction unit, a spatial frequency detection unit that detects a spatial frequency, a specific color detection unit that detects the presence of a specific color from the values of the color difference signals Cr and Cb, and the spatial frequency A counter that concentrates on a pixel having a specific value and a counter selection unit that selects the counter that concentrates the specific color, and the correction curve generation unit includes the counter selection unit. An image quality comprising: a restriction processing unit that adjusts a count value from a counter that is selected as a counter in which a spatial frequency is concentrated on a pixel having a specific value and is selected as a counter in which a specific color is concentrated Correction circuit.   In the restriction processing unit, the counter selection unit adjusts the count value from the counter selected as a counter in which pixels with low spatial frequency values are concentrated and selected as a counter in which skin color as a specific color is concentrated. 6. The image quality correction circuit according to claim 5, wherein the image quality correction circuit is performed.   A matrix conversion processing unit is provided after the gradation characteristic correction unit, and the matrix conversion processing unit detects a correction ratio from the input luminance signal Y and the corrected luminance signal Y ′ from the gradation characteristic correction unit. After correcting the color difference signals Cr and Cb at the same rate as this correction rate, the matrix conversion processing is performed using the corrected luminance signal Y ′ and the corrected color difference signals Cr and Cb. 6. The image quality correction circuit according to claim 1, 3, or 5.   A matrix conversion processing unit and a γ correction processing unit are sequentially provided after the gradation characteristic correction unit. Among these, the γ correction processing unit detects an image having a high spatial frequency and a high gradation detected by the spatial frequency detection unit. 6. The image quality correction circuit according to claim 1, wherein a γ correction process is performed by selecting a γ correction curve so that a high gradation portion is not crushed.

Images